Host command execution via unauthenticated MCP configuration
Published Jul 15, 2026 · Updated Jul 15, 2026
Command injection in GPT Researcher 3.3.7 through 3.4.4 allows remote attackers to execute host commands through its WebSocket endpoint. The unauthenticated /ws handler passes attacker-controlled MCP command, argument, and environment values to anyio.open_process without validation or an allowlist. Direct exploitation requires network access to the service; a crafted page can reach a normally local instance when a victim opens it.
Summary
What happened
Command injection in GPT Researcher 3.3.7 through 3.4.4 allows remote attackers to execute host commands through its WebSocket endpoint. The unauthenticated /ws handler passes attacker-controlled MCP command, argument, and environment values to anyio.open_process without validation or an allowlist. Direct exploitation requires network access to the service; a crafted page can reach a normally local instance when a victim opens it.
The record
- CVE
- CVE-2025-65720
- Published
- Jul 15, 2026
- Updated
- Jul 15, 2026
- Vendor
- Unknown vendor
- Product
- Unknown product
- Classifications
- T1190, T1059
- Attack vector
- network
- Privileges
- unauthenticated
Timeline
How it unfolded
- Jul 15, 2026CVE publishedPublication date reported by the CVE source.
- Jul 15, 2026Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
What conditions does exploitation require?
What is affected?
Affected products and versions are unavailable in this record.
Published CVSS scores
CVSS describes severity. EPSS estimates exploitation probability.
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Weakness, pattern, technique
Public exploit references
- GPT Researcher WebSocket MCP command-execution proof of conceptfunctional · demonstrated
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Book a demo to see how Hinoki identifies affected software and validates exploitability in your environment.
Book a demo